A Bayesian model of fisheries discards with flexible structure andpriors defined by experts

tMinimizing the probability of discards is an important step in mitigating environmental impacts of fishingactivities and maximizing economic gains from fish stocks. Although several discard models have beenrecently developed, current approaches are still unable to robustly adjust to different circumstances (e.g.fishery target species, geographical location), and lack a closer interaction with stakeholders. Here, wepresent a novel approach consisting of a modular Bayesian model, which can incorporate any relevant explanatory variable, independently of data availability. The relationships between the variables anddiscard rates are initially delineated by stakeholders through surveys, and included to the model aspriors. The priors are then used together with observed data for estimating posterior distributions ofdiscard probability. We test this approach in two study areas at the Mediterranean Sea: the Ligurian andTyrrhenian Seas in Italy, and the Greek part of the Aegean Sea. For each site, we evaluated the modelfor estimating discards from bottom trawl fishery associated with European minimum conservationreference size regulations, as well as discards caused by the low economic value of catches